Properties of the time-optimal feedback control for a pendulum-like system (Q467483)

The paper follows up previous works by the authors on the structure of the minimum-time optimal control for nonlinear pendulum systems. These systems are described by second-order ordinary differential equations with a nonlinear periodic forcing term whose value and time derivative are assumed to be bounded. The scalar control compensates for the forcing term, and it is assumed to be bounded as well. For these problems, the optimal control is known to be bang-bang, i.e. it takes only extreme values. Using the classical Pontryagin Maximum Principle (characterizing the first-order necessary condition for optimality), the authors study comprehensively the geometry of the switching curves, consisting of points in the two-dimensional state-space (angle and angular velocity) at which the optimal control changes sign, and the so-called dispersal curves, consisting of points at which two distinct optimal trajectories are meeting. The authors also derive explicit estimates on the control norm such that the number of optimal switchings does not exceed one for any initial condition.